Title: Is rETH’s fee costing it demand? A data study: plan, first numbers, feedback welcome
Hi all, I’m Nishikawa. I’m building an open, reproducible study of a question the June voter-share thread raised but could not settle with data: how much does rETH demand respond to its yield gap against stETH, and what would a lower fee do for holders, node operators and RPL voters?
Why now. rETH supply fell 45% since January 2024, from 562,350 to 309,436 on 24 October 2026. The June thread had strong arguments on both sides, but no time series of rETH’s yield against its competitors and no estimate of how flows respond.
Three questions.
- How much less, or more, does rETH pay its holders than stETH, week by week since January 2024? Measured from each token’s exchange rate, so fees are already inside.
- Do weekly rETH flows follow that gap, and by how much?
- What would 14%, 12% and 10% fees do over 12 months? One number frames it: cutting the voter share from 9% to 5% keeps RPL voter income flat only if rETH grows 80%.
First numbers.
- Average yield gap since January 2024: -29.7 basis points, with rETH behind stETH in 93% of weeks.
- Average gap since Saturn 1: -25.5 basis points. In the third quarter of 2026 it shrank to 3 basis points, yet rETH kept shrinking while stETH grew.
- rETH’s market price against its protocol rate: an average discount of 0.13%.
Preliminary model results, open for review. Weekly flows barely respond to the gap: the slope is indistinguishable from zero at one to three weeks, and small at four, which is not significant once the four lags tested are taken into account. At that response, a 12% fee would leave RPL voter income about 22% lower after a year, and breaking even would need a response about six times larger than the data allows. The result holds with stETH’s flows as a market-wide control, and the decline itself is rETH-specific: over the same weeks, stETH’s net flows came to -3.5% against rETH’s -45%. The data cannot show what happens if a cut lifts rETH above stETH, because rETH sat below stETH in 93% of weeks.
Method and code. Public data only: the chain for rates and supply, DefiLlama for prices and block numbers. The notebook reruns every week on GitHub Actions, and an offline test checks that the method recovers a known response on a synthetic market: GitHub - NishikawaButterfly/reth-competitiveness-study · GitHub
What would help.
- Prior work or data I should use and credit, such as the analysis behind RPIP-73’s threshold.
- Drivers of flows I am missing: queue lengths, integrations, campaigns.
- Whether modelling the fee change as reaching only megapool ETH is the right assumption.
The full write-up, with your feedback folded in, will follow by the end of November.
Disclosure: written with AI assistance (Claude). I hold no RPL and no rETH.